Bead shape in high frequency electric resistance (HER) pipe welding gives useful information on judging current welding condition. In most manufacturing process, heat input is controlled by skilled operators observing color and shape of bead. We designed a visual monitoring system of bead shape in HERW pipe process by using structured light beam. To observe the bead shape, we project a slit light on the bead in an oblique angle, and take a light sectioning profile by C.I.D. camera. There are some difficulties in finding the relationship between the bead profile and heat input to be controlled, because this relationship is not one-to-one and the bead profile is easily contaminated by environment noise. To get over such difficulties, Kohonen Network is used to classify bead shapes and rule-based heat input classifier is designed to relate heat input and bead classification result. The experimental results show superb performance of the proposed method.
